High alpine sorcerers: revision of the cave wētā genus Pharmacus Pictet & de Saussure (Orthoptera: Rhaphidophoridae: Macropathinae), with the description of six new species and three new subspecies

The New Zealand alpine cave wētā genus Pharmacus was first described by Pictet & de Saussure (1893) as a monotypic taxon. Three species were added to the genus by Richards in 1972. Here we clarify the status and appearance of all known species of Pharmacus. Based on morphology and mtDNA sequences we determine that the species Pharmacus brewsterensis Richards, 1972 is better placed within the genus Notoplectron Richards, 1964. We also resolve the species Isoplectron cochleatum Karny, 1935 and show that it belongs to the genus Pharmacus. Additionally, we describe six new species and three new subspecies from the southern regions of South Island, New Zealand. We provide key traits and known distributions for all known species and subspecies in this alpine genus. New combinations: Pharmacus brewsterensis Richards, 1972 becomes Notoplectron brewsterense (Richards, 1972) comb. nov.; Isoplectron cochleatum Karny, 1935 becomes Pharmacus cochleatus (Karny, 1935) comb. nov. New species and subspecies: Pharmacus cochleatus rawhiti subsp. nov., Pharmacus cochleatus fiordensis subsp. nov., Pharmacus cochleatus nauclerus subsp. nov., Pharmacus concinnus sp. nov., Pharmacus cristatus sp. nov., Pharmacus notabilis sp. nov., Pharmacus perfidus sp. nov., Pharmacus senex sp. nov. and Pharmacus vallestris sp. nov. New synonyms: Pharmacus dumbletoni Richards, 1972 = Pharmacus montanus Pictet & de Saussure, 1893 syn. nov.; Pharmacus chapmanae Richards, 1972 = Pharmacus cochleatus (Karny, 1935) syn. nov.


Introduction
Cave wētā (Family Rhaphidophoridae Walker, 1869) are common and widespread in the alpine regions of New Zealand, where they hide in rock crevices and in holes under boulders, in landscapes that are covered by snow for more than six months a year, and survive at elevations as high as 2800 m a.s.l. (Pictet & de Saussure 1893;Dumbleton 1935;Richards 1972;Walker 1977;Sweney 1980). Sampling in the alpine regions is made difficult by the rugged terrain and the short summer seasons, which is probably why studies of alpine Rhaphidophoridae in New Zealand have been few and far between. Not only is the list of publications limited to the few titles listed above, but most species were described based on a very small number of individuals.
The alpine cave wētā genus Pharmacus was established by Pictet & de Saussure (1893) as a monotypic genus, with Pharmacus montanus as type species, based on the examination of one adult male specimen. For nearly 80 years, this remained the only New Zealand alpine rhaphidophorid known to science, with the common name 'Mount Cook flea'. A re-description of the genus Pharmacus by Hutton (1896) did not add anything, since Hutton did not acquire any new material, nor did he examine any specimens, but relied on verbal information supplied by Pictet. A revision of the genus Pharmacus was completed by Richards (1972), who re-described Pharmacus montanus after examination of additional material, and designated three new species: Pharmacus brewsterensis, P. chapmanae and P. dumbletoni. Richards (1972) was able to secure a good number of male and female specimens of Pharmacus montanus and of her new species P. chapmanae. However, she described P. dumbletoni and P. brewsterensis based on one male and two females respectively.
The possible existence of additional species of Pharmacus was hinted at by a number of authors but has never been formally resolved. In her revision of Pharmacus, Richards (1972) wrote "A possible fifth species occurs in the Diorite Range, northern Fiordland. It is closely related to P. chapmanae, but is known only from one damaged adult female. More specimens are required before its true affinities can be determined". A Diorite Range does not exist in Fiordland, nor anywhere in New Zealand. Based on alliteration and geographical placement, we can only assume that Richards was referring to Fiordland's Darran Mountains. A new species of Pharmacus is also listed as having been collected on Way Spur (Symmetry Peaks, Eyre Mountains) during an ecological survey of the Eyre Ecological District (Mark et al. 1989). The whereabouts of the materials collected during the survey is unknown. Two putative new species of alpine cave wētā from the Remarkables (Wakatipu) and Gertrude Saddle (Darran Mountains) are pictured in the popular book "Which New Zealand Insect?" (Crowe 2002); both clearly belong to the genus Pharmacus. Richards (1972) assigned the two female type specimens of Pharmacus brewsterensis to the genus Pharmacus based on these having been collected in a high alpine environment, on them being armed with one retro-lateral apical spine at the apex of the mid femur, and on the smooth upper valve of their ovipositors. Female rhaphidophorids are more difficult to classify since they do not show the same level of variation in the terminalia as males do (Hegg et al. 2019). It does not help that Richards did not collect any of her alpine specimens and had to rely for her descriptions on material collected by others, often poorly preserved. Considering these limitations, it does not surprise that she could have made mistakes in her taxonomy.
The taxonomy of New Zealand Rhaphidophoridae is further complicated by the fact that several species were described in German or French, and that translations of the original descriptions have never been published. The original description of Pharmacus montanus for instance is in French. Richards (1972) assumed the description provided by Hutton (1896) to be a faithful translation of the original. However, this is not the case, and important detail was lost in Hutton's rendition. Likewise, the original description of Isoplectron cochleatum Karny, 1935 is a wordy and poorly illustrated three pages written in German.
As a result, the species Isoplectron cochleatum has remained obscure to most and has been ignored by entomologists until present, as shown by its complete absence in the literature on New Zealand Rhaphidophoridae. To remedy this, faithful translations of the original descriptions of Pharmacus montanus and of Isoplectron cochleatum are published in this paper's Supp. file 1: Appendix A and B.
Here, we review the status of the alpine cave wētā genus Pharmacus using morphological and genetic data from a large sample. We identify male Pharmacus brewsterensis for the first time, since no males were included in the original description. We compare Pharmacus brewsterensis with other species in the genus Pharmacus and in other genera of New Zealand Rhaphidophoridae, to test its placement in Pharmacus. After examination of the type material, we resolve the species Isoplectron cochleatum and we revise its taxonomic status. We use a combination of dense sampling throughout the South Island of New Zealand's alpine regions, morphological traits and DNA sequence data to test the status of putative Pharmacus taxa. We use morphological traits to identify six hitherto undescribed species and three new subspecies within this genus. We also provide new descriptions for previously described species, to account for new synonymies and for the variability of traits in specimens we have collected over a much wider geographical range compared to previous studies.

Collection and morphological methods
Cave wētā were collected opportunistically around New Zealand, including at the holotype location of all known species, using day and night searching of cliffs, mountain ridges, rock tors, forests and caves, and occasional pitfall trapping. More than 5200 specimens have been catalogued; many have been examined in detail and sampled for DNA sequence comparison, among these 373 specimens of Pharmacus (including Isoplectron cochleatum and excluding Pharmacus brewsterensis). Specimens are held in the Phoenix Lab collection at Massey University (MPN), except for type material, which is lodged at Museum of New Zealand Te Papa Tongarewa (NMNZ). Pharmacus specimens were identified based on the descriptions by Pictet & de Saussure (1893), Hutton (1896), Karny (1935) and Richards (1972).
Specimens were examined and photographed using a DSLR camera (Nikon D800, Nikon D850 or Sony α7RII) attached to a Nikon Plan 4/0.13 microscope tip and Nikon PB-6 bellows, mounted on a Cognisys Stackshot 3× automated rail. Focus stacks were generated using the software Helicon Focus ver. 6.8.0 Pro (Helicon Soft Ltd 2000). Adults were distinguished from immature individuals by darker, sclerotised bodies and fully formed external genital structures. In particular, the pigmentation, shape and sharpness of ovipositors, subgenital plates and cerci were informative about developmental stage. We looked for the presence/absence of each of 22 apical leg spines (Fitness et al. 2015) (Supp. file 1: Fig. S1), as well as the combinations and numbers of linear spines on the legs, and the shape of the subgenital and suranal plates.
Measurements of key body parts were obtained using digital callipers (Table 1). All linear measurements and count data were tested for sexual dimorphism using Linear Mixed Models in JASP ver. 0.14.1 (JASP Team 2020), with measurement as the response variable, sex and species as fixed factors and geographical location as random factor. Where sexual dimorphism is detected and is statistically significant, measurements are reported separately for males and females; otherwise the measurements for both sexes are pooled. Some traits included in Table 1 are invariable in Pharmacus (e.g., teeth on dorsal valve of ovipositor; pairs of longer spines on hind tibia) -they are included nonetheless since they are useful for comparison with other genera of NZ Rhaphidophoridae (see for example Fitness et al. 2018:  Two-letter codes in the 'Material examined' sections below refer to the New Zealand entomological regions (Crosby et al. 1998) (Fig. 1B).

Molecular methods
Genomic DNA was extracted from leg tissue of specimens representing each morphotype, using a salting-out protocol (Trewick & Morgan-Richards 2005). For most samples, a ~1500 base pair (bp) fragment spanning most of the mitochondrial cytochrome c oxidase I (COI) gene of the mitochondrial genome was amplified using polymerase chain reaction (PCR) with the invertebrate primers: LCO1490 (Folmer et al. 1994) and L2-N-3014 (Simon et al. 1994). Where DNA was of lower quality a shorter fragment (~800 bp) was amplified using primers C1-J-2195 and L2-N-3014 (Simon et al. 1994).
Successful PCR products were sequenced with Bigdye chemistry on an ABI 3730 genetic analyser (Applied Biosystems Inc., Carlsbad, CA) using primer L2-N-3014 at the 3' end of COI. A subset of 14 specimens were sequenced with LCO1490 (Folmer et al. 1994) and L2-N-3014 (Simon et al. 1994) to provide sequences spanning COI. Nucleotide sequences were assembled and aligned using Geneious Prime ver. 2020.2.2 (https://www.geneious.com; Kearse et al. 2012). No insertions/deletions were detected and sequences were translated to confirm that there were no stop codons or frame shifts that would indicate the presence of nuclear paralogs. A selection of 60 sequences was submitted to GenBank; accession numbers (OM293676-OM293737) are listed in the Material examined section.
We examined the relationships of putative Pharmacus taxa by phylogenetic reconstruction using Maximum Likelihood criteria applying a GTR evolutionary model with a gamma-distributed rate variation across DNA sites and a proportion of invariable sites. This was implemented with the PhyML plugin (Guindon et al. 2010) in Geneious Prime. Representative, homologous mtDNA COI sequences from other New Zealand cave wētā: Talitropsis sedilloti Bolívar, 1882 (MPN CW1830), Setascutum pallidum Richards, 1972 (CW3158) and Notoplectron campbellense Richards, 1964 were used for comparison; Macropathus filifer Walker, 1869 (CW226B) was included as an outgroup based on available phylogenetic information (Allegrucci et al. 2010;Fitness et al. 2018;Hegg et al. 2019) and unpublished analyses. A reduced taxon dataset with just 14 putative Pharmacus specimens (including P. brewsterensis) allowed Maximum Likelihood inference with 1000 bootstrap replicates to assess monophyly of the genus. We identified adult specimens using morphology and used mitochondrial DNA sequences to test these species hypotheses, to confirm association of males and females within each taxon, and to identify nymphs, which are often too difficult to identify to species level based on morphology alone. Maximum Likelihood was then used with mtDNA sequence from 60 putative Pharmacus specimens to infer evolution relationships of the haplotypes to illustrate the concordance between morphologically distinct taxa and unique haplotypes.
Mitochondrial DNA sequence variation allowed us to verify prior hypotheses of distinct taxa based on morphological traits and to seek concordance with the newly morphologically identified taxa in this genus. The correlation of morphological and genetic clusters is expected from distinct evolutionary  (Crosby et al. 1998). These codes are reported in the Material examined section for each species.
lineages that are consistent with their treatment as distinct taxonomic units (Mallet 1995(Mallet , 2013b. We note that large stable populations observed in New Zealand insect species tend to yield distributions of pairwise mtDNA differences that deviate from the expected exponential distribution owing to their common history (Slatkin & Hudson 1991;Morgan-Richards et al. 2017), which can mislead species delimitation tools that rely on this single non-recombining locus (Dellicour & Flot 2015). In addition, lack of lineage sorting, selection on morphology and hybridisation can all result in gene trees differing from species trees. Therefore, we were not attempting to resolve a fully supported phylogenetic hypothesis, simply illustrating the concordance between haplotypes and phenotype.

Identity of previously described species
Using the morphology of terminalia in adult specimens and the key provided by Richards (1972), we identified cave wētā that could be assigned to all known species of Pharmacus, and six other related and as yet undescribed species. Our phylogenetic analysis of mtDNA sequences places Pharmacus brewsterensis outside the Pharmacus clade, sister to Notoplectron campbellense from the subantarctic Campbell Islands (Fig. 1A). This result was supported in 1000 out of 1000 bootstrap resamples of our data. Five of the six new species identified by morphology were each found to correspond to a distinct mtDNA lineage within the monophyletic Pharmacus clade (Fig. 2). Specimens of a morphologically distinct taxon from Fiordland (P. cristatus sp. nov.) had mtDNA haplotypes that nested within the diversity of widespread P. cochleatus (Karny, 1935) comb. nov. Due to the wide distribution and colour and spine variation detected within P. cochleatus we analysed COI sequence variation using an unrooted network approach (not shown); however, we could not resolve two of the four new subspecies of P. cochleatus using just mtDNA.
We found Pictet & de Saussure's (1893) and Richard's (1972) descriptions of Pharmacus montanus to be accurate. The species is readily identified based on the shape of the male terminalia ( Fig. 3A-E) and the lack of dorsal spines on the first hind tarsus segment. One interesting piece of information present in Pictet & de Saussure's (1893) original description in French is the statement "The apex of the hind tibiae is armed below with three small spines, two of which are on the inner edge" (see the translation in Supp. file 1: Appendix A). This detail was omitted in Hutton's (1896) translation and subsequently by Richards (1972). We have found this trait in two of five males we examined from the Sealy Range near Aoraki/Mount Cook Village ( Fig. 3F-G), but not in any specimens collected elsewhere. We know that the Pharmacus montanus holotype location is somewhere near Aoraki/Mt Cook at 7000 ft of elevation (Pictet & de Saussure 1893). It is possible that the holotype of P. montanus may have originated from Mt Annette, a readily accessible mountain that is popular with climbers, at an elevation of 7000 ft in the Sealy Range. Given our small sample size however this remains little more than an educated guess.
After collecting several specimens of Pharmacus dumbletoni near the holotype location on Gloriana Peak, Spenser Mountains, we found that we could not tell these apart from Pharmacus montanus morphologically or genetically (see red frame in Fig. 2). The colour pattern of the tergites and the shape of the male genitalia are the same. Richards (1972) judged one male specimen from the Spenser Mountains to belong to a new species (P. dumbletoni) based on the lack of ventral linear spines on the hind femora. While P. montanus mostly has armed hind femora in the Mount Cook Region and unarmed hind femora in the Spenser Mountains, we did collect specimens with unarmed hind femora in the Mt Cook Region, and specimens with armed hind femora in the Spenser Mountains. We conclude that the number of ventral linear spines on the hind femora in P. montanus is subject to both regional and individual variation and is not characteristic of the species. This is consistent with what had already been observed in other taxa of New Zealand Rhaphidophoridae (Fitness et al. 2015;Hegg et al. 2019). P. dumbletoni Richards, 1972 is thus a junior synonym of P. montanus Pictet & de Saussure, 1893. While Richards (1972) had examined several specimens of Pharmacus montanus collected in Canterbury between Aoraki/Mount Cook Village and the Craigieburn Range, we found the species as far north as Mount Owen in Kahurangi National Park (see Fig. 18A). The variability of the species over such a wide geographical range and the synonymy of Pharmacus dumbletoni warrant a re-description of P. montanus.
We collected several specimens of Pharmacus brewsterensis on Mt Brewster (holotype location) and elsewhere and identified males for the first time, since the species was described based only on two females. In her species description Richards (1972) noted several differences compared to other species of Pharmacus, namely the lighter overall coloration, the presence of fewer teeth on the ovipositor, the greater number of linear spines on the hind femur, and the shape of the female subgenital plate. The latter has two acute, pointed lobes, as opposed to the rounded lobes found in all other species of Pharmacus ( Fig. 4D-G). Richards described the hind tibiae of P. brewsterensis as being "armed above with 11 to 13 prolateral and 12 to 13 retrolateral linear spines", but did not comment on how these numbers are only one half to two thirds of the number of dorsal linear spines on the hind tibiae in all other species of Pharmacus (see Table 1, Fig. 4H-I). Neither did she note that the dorsal linear spines on the hind tibiae of P. brewsterensis are visibly socketed at the base, whereas they are fused to the shaft of the tibia in all other species of Pharmacus ( Fig. 4J-K). Most importantly, Richards did not have a chance to examine the structure of the male terminalia in P. brewsterensis, which is entirely different from all other Pharmacus, without a subgenital plate, the paraprocts very prominent at the apex of the body ( Fig. 4A-C). Our mtDNA sequence analysis (Fig. 1A) shows that Pharmacus brewsterensis is more closely related to Notoplectron campbellense Richards, 1964 than to any of the Pharmacus species. We suggest that this species be transferred to the genus Notoplectron Richards, 1964 as Notoplectron brewsterense (Richards, 1972) comb. nov. Further discussion of this species awaits a revision of the genus Notoplectron.
We identified Pharmacus chapmanae based on the original description and the shape of the male and female terminalia ( Fig. 5C-F). We also examined the type material held at the Otago Museum in Dunedin, New Zealand (Supp. file 1: Fig. S4). The holotype's (male) sub-genital plate was removed by Richards and was not preserved with the specimen. This makes the holotype of limited use for identification to species level. Richard's drawings remain our best diagnostic tool ( Fig. 5C, E), and are in good agreement with the terminalia of an alpine species that is both common and widespread in the Southern Alps south of Aoraki/Mount Cook (Fig. 5D, F). An examination of the holotype of Isoplectron cochleatum reveals that this is the same species (see Fig. 5A-B). Having identified all known species of New Zealand Rhaphidophoridae, we can confidently state that there is no other species of cave wētā this could be confused with. Since Isoplectron cochleatum was described first, the name takes precedence. At the same time, the species belongs to the genus Pharmacus Pictet & de Saussure, 1893, not to Isoplectron Hutton, 1896 (see Fig. 2). We thus designate the new combination Pharmacus cochleatus (Karny, 1935) comb. nov., with Pharmacus chapmanae Richards, 1972 as a junior synonym.
It is worth noting that the allotype (MNHN EO-ENSIF4927) and the female paratype (MNHN EO-ENSIF4929) of Isoplectron cochleatum belong to a different species. The allotype is a female nymph, and the only one of four type specimens to have retained its hind legs. The dorsal spines on the hind tibiae suggest this specimen belongs to the species Talitropsis sedilloti. Identification of the female paratype is more difficult, but the shape of the sub-genital plate suggests this is likely to also be Talitropsis sedilloti. The insect in Karny's (1935)  The labels for the holotype (MNHN EO-ENSIF4926) and the male paratype (MNHN EO-ENSIF4928) of Pharmacus cochleatus comb. nov. state that these specimens were collected in Nelson, New Zealand, in 1876, sixty years before the species was described. We are confident that the collection locality was recorded incorrectly, and that the specimens originate from the alpine regions in the south of New Zealand's South Island. This is not surprising, as other studies have highlighted how the location of origin of specimens in museum collections is often recorded incorrectly (Boessenkool et al. 2010;Verry et al. 2019).
While the exact origin of Karny's type specimens remains unknown, Richards (1972) obtained several specimens from the Otago Alps for her description of Pharmacus chapmanae. We have extended the known distribution range of Pharmacus cochleatus comb. nov. significantly, having examined specimens from a much wider spread of locations including northern Fiordland, Central Otago and the Southern Alps south of Aoraki/Mt Cook (see Fig. 18B).
Specimens of Pharmacus cochleatus comb. nov. from the Southern Alps and Fiordland are morphologically identical, yet genetically they fall into three clusters (Fig. 2). Nested within the mtDNA diversity of Pharmacus cochleatus is the clade of the morphologically distinct Pharmacus cristatus sp. nov. found sympatric with P. cochleatus but with unique male terminalia (Figs 2, 10). The central-northern clade of P. cochleatus is separated from the southern (Fiordland) clade by the northern arm of Lake Wakatipu and the valley of the Dart River (Figs 6, 18B). The third clade is endemic to the Skippers Range in South Westland, an isolated mountain range that reaches an elevation of 1650 m a.s.l. and is surrounded by valleys and passes that are entirely below 50 m a.s.l. (Fig. 6). We suggest that that the concordance of geographical separation and clustering of mtDNA haplotypes warrants these three populations being assigned to different subspecies (Cronin et al. 1996;Miller et al. 2011). A difficulty arises in that we do not know where the holotype of Pharmacus cochleatus was collected in 1876. Morphologically, it could belong to any of the three groups. Extracting DNA from a 145 year old, dry holotype specimen does not seem like a realistic proposition, either. The best we can do is to infer from the history of exploration of New Zealand's southern regions. While the exploration of coastal Fiordland started in the 18 th century, the first expedition into the interior where any mountain tops were climbed was by surveyors James McKerrow and Goldie in 1862/63 (Hall-Jones 1976). This was a two-man expedition with no scientists on board; no scientific samples were collected. The mountain ranges between Lake Te Anau and Milford Sound were not crossed until 1888, and the exploration of Fiordland's alpine regions did not start in earnest until the turn of the 20 th century (Hall-Jones 1976 A further complication arises in a population of Pharmacus cochleatus comb. nov. from the mountain ranges either side of Lake Wakatipu and south of the Greenstone and Kawarau Rivers (see Fig. 6), inhabiting two alpine islands that are geographically disconnected from the species' main distribution range. Males and females from this population have the same terminalia as all other Pharmacus cochleatus examined and are not genetically partitioned from Pharmacus cochleatus fiordensis subsp. nov. Nevertheless this eastern population is readily distinguished by body-and eye-colour, different spines at the apex of fore and mid femur, and longer legs in proportion to the rest of the body (see Table 1). We suggest that this population also fits the definition of a subspecies in that it inhabits a clearly distinct and separated part of the species' range and is morphologically notably different (Monroe 1982;Mallet 2013a); we propose the name Pharmacus cochleatus rawhiti subsp. nov.  Table S12. Richards (1972) did examine Pharmacus montanus from Hamilton Peak in the Craigieburn Range, which means the species is most likely to be present there even though we did not find it. Meads & Notman (1992) listed Pharmacus montanus among their 'incidental invertebrates observed during surveys' for giant wētā (Deinacrida sp.) in the Inland and Seaward Kaikōura Ranges. Meads & Notman did not give locations, nor did they state whether they collected any specimens. While their identification appears dubious, the Kaikōura Ranges could be a worthy target of more intensive searches in the future.
2 Body length is measured from the apex of the fastigium to the posterior margin of the suranal plate.

Description
A genus of small to mid-sized cave wētā (adult body length 10 to 18 mm, typically around 15 mm) with a distribution limited to the high alpine regions of the South Island of New Zealand. Only found above tree-line, often above snow-line, at elevations as high as 2800 m a.s.l.
All species in the genus look similar to each other. While this makes it easy to characterise the genus, an examination of the adult male terminalia is often required for a reliable identification to the species level. A detailed description of the genus follows below; individual species descriptions focus on those traits that vary between species or that differ from the generic description.
Head. Oval in shape; glabrous except for maxillary palps and antennae. Frons with a pale median line and two dark stripes converging to the fastigium on either side (Fig. 7A, C). Eyes rounded, but with a straight inner edge facing the fastigium. Face generally mottled with dark and pale patches. Fastigium pointed and ending abruptly below; dark with pale patches on either side; divided by a deep median groove. Scapes of antennae approximately three times as long as pedicel. All parts of the antennae pale or reddish-brown and covered with fine setae, except for the rotating joints. No visible sexual dimorphism in scapes of antennae or any other head-part (Fig. 7). Maxillary palps pale, of varying length, with moderately dense covering of hair.
THorax. Colour and tomentum on pronotum variable both between and within species. A pale, thin median line is generally present, although it may be inconspicuous (see Fig. 8). Lateral edges of pronotum with a pronounced rim and bent upwards in all species.
Legs. Moderately long. Hind femora generally shorter than body; hind tibiae between 30% shorter and 10% longer than body in females, between 20% shorter and 20% longer than body in males. Coxae and trochanters mottled light and dark brown. Fore and mid femora and tibiae may be uniform pale or brown, or variegated; hind legs variegated. Fore coxae with a pronounced lateral anterior spine. Fore femora without linear spines above or below, but may be armed with one prolateral spine at the apex in some species. Fore tibiae armed below, generally with two linear spines on both anterior and posterior edge in all species. Fore tibiae armed with two long spines below (one prolateral and one retrolateral) and two short spines above (one prolateral and one retrolateral) at apex. Mid femora without linear spines above or below, but always armed with one retrolateral spine at apex. A prolateral spine at the apex of the mid femur may be present in some species. Mid tibiae armed below, generally with two linear spines on both anterior and posterior edge in all species. Dorsal linear spines on the mid tibiae are rare but possible. Mid tibiae armed with two long spines below (one prolateral and one retrolateral) and two short spines above (one prolateral and one retrolateral) at apex. Hind femora may be armed with linear spines below, but never with apical spines. Hind tibiae armed with an average of 22 linear spines above, of varying length, on both anterior and posterior edges (Fig. 9). The spines are fused to the shaft of the tibia and are not socketed or articulated (Fig. 4K). Hind tibiae with two superior subapical spines (one prolateral and one retrolateral), two superior apical spines (one prolateral and one retrolateral), two inferior apical spines (one prolateral and one retrolateral) and two inferior subapical spines (one prolateral and one retrolateral). Superior apical spines approx. 50% longer than inferior apical spines; inferior apical spines approx. 50% longer than superior subapical spines; superior subapical spines approx. 50% longer than inferior subapical spines. Hind tarsi with four segments; first and second segments with a pair of spines on distal end. First segment and occasionally second segment armed above with small, alternate dorsal spines, in all species except Pharmacus montanus. The length of the tarsus segments (ordered from first to fourth) is in a ratio of 9 : 3 : 1 : 5.
abdomen. Colour of tergites typically chequered, alternating pale and dark patches (Fig. 8). A uniform brown colour is common in adult Pharmacus montanus and in the nymphs of several species; a uniform black colour is common in both P. montanus and P. cochleatus populations above the permanent snowline. Dorsal median line generally visible but may be thin and inconspicuous. Tomentum variable both between and within species.
maLe TerminaLia. Cerci between 15% and 20% of body length, pointed at apex, variable in colour, clothed in setae. The distal half is often visibly thicker than the proximal half (see Figs 5A-B, 10E, K, 11E, G-H). Styli short and stumpy, covered in sparse setae. The subgenital plate looks fairly similar in all species, broadly triangular, but enlarged in a spoon-like structure at the apex. The difference between species however is pronounced enough to provide the strongest character for species level identification (see Figs 10-11).
FemaLe TerminaLia. Subgenital plate bilobed and rounded in all species. Ovipositor reddish-brown, straight or gently curved upwards at apex, terminating in a sharp point; relatively short (approximately two thirds of body length). Upper valve always smooth above; lower valve with 5 to 9 strong teeth at apex on ventral edge (Figs 12-13).
Nymphs generally look the same as adults, only smaller, which makes them readily identifiable as Pharmacus.

Diagnosis
A mid-sized cave wētā found in the Southern Alps from the Ben Ōhau Range northwards, at elevations greater than 1200 m. Body colour varies from marbled brown to black, generally with a thin but well visible median dorsal line along the whole length of the animal, and pale first tarsal segments. The pronotum and tergites are always covered in fine tomentum, giving the animal a matt appearance.
At the southern end of its distribution range, Pharmacus montanus overlaps with P. cochleatus comb. nov. and could be easily confused with the latter, especially at the higher elevations, where both insects are black. P. cochleatus is often glabrous and shiny in appearance and has first segment of the hind tarsi armed with dorsal linear spines, which are absent in P. montanus.
On the higher mountain tops east of the Main Divide of the Southern Alps, P. montanus shares the habitat with Petrotettix serratus Richards, 1972. The latter species is often dark with pale tarsi like P. montanus but has dorsal linear spines on the first two segments of the hind tarsi, and a serrated upper valve of the ovipositor. In the alpine regions around Mt Cook, P. montanus forms mixed populations with Notoplectron brewsterense comb. nov. The latter is yellow-orange in colour, has a stocky appearance due to its shorter legs, and has fewer, larger dorsal linear spines on the hind tibiae. See also Fig. 4 for a comparison between the two species.
At the northern end of its distribution range in Kahurangi National Park, P. montanus is uniform brown and shares the habitat with Macropathus filifer, which is also brown. The latter species however has much longer legs, which give it a very slender appearance.
In common English, P. montanus is also known as 'Mount Cook flea'. The name is explained by the insect's "reprehensible habit of leaping in showers out of a crevice in the rock upon the unsuspecting climber" (Dumbleton 1935), and is still in common use today.
Material examined (see also Supp. file 1: Head. As per generic description. Specimens living above the permanent snowline are entirely black. This includes all head parts that are usually pale otherwise, e.g., maxillary palps and antennae.
THorax. As per generic description; always covered in fine hair, which gives the animal a matt look.
Legs. Longer than in most Pharmacus species. The hind tibiae are 5% longer than the body in females, 20% longer than the body in males. Fore femora always unarmed at the apex. Mid femora armed with one retrolateral spine at the apex; prolateral apical spine absent. Hind femora occasionally armed with up to two linear spines below, on either or both inner and outer edge. Hind tibiae armed with two superior subapical spines (one prolateral and one retrolateral), two superior apical spines (one prolateral and one retrolateral), two inferior apical spines (one prolateral and one retrolateral) and two inferior subapical spines (one prolateral and one retrolateral), as in all other Pharmacus species. However, in some specimens in the Mt Cook region, a third inferior subapical spine may be present on the inner edge of the tibia (see Fig. 3F-G). First and second tarsal segments without any linear spines above -a trait that differentiates P. montanus from all other Pharmacus species. The first tarsal segment is generally pale on all six legs, even in specimens that are entirely black otherwise -the contrast is in fact especially noticeable in the darker specimens and gives the insect a characteristic look (Fig. 14A).
abdomen. Tergites always covered in fine hair, which gives the animal a matt look. A thin, pale median line along the whole length of the insect is generally visible, even in the darker specimens. The colour of the tergites is most commonly brown at the lower elevations (Fig. 14B), transitioning to black at the higher elevations, especially above the permanent snow-line (Figs 8A, 14A). Specimens in the Arthur's Pass region are chequered, unlike other P. montanus but like other Pharmacus species further south (Fig. 8B).
maLe TerminaLia. Subgenital plate triangular when seen from below; deeper than in all other Pharmacus species when seen from the side (Fig. 10A-C). In a dorsal view, a thin plate or membrane covered in tomentum is attached to the vertex of the triangle. The plate varies in shape from oval (Figs 3D, 8B) to rectangular with rounded corners (Figs 3E, 10A).
FemaLe TerminaLia. Subgenital plate strongly bilobed, the two rounded lobes long and narrow, separated by a deep V-shaped depression; hairy (Figs 4F-G, 12A). Ovipositor on average 70% of body length, relatively straight, tapering gently at first then more strongly near the apex; lower valve with 5 to 8 strong teeth below at the apex ( Fig. 12B-C). (Karny, 1935) comb. nov. The species, originally described by Karny (1935) as Isoplectron cochleatum based on two male specimens of unknown origin, was described a second time as Pharmacus chapmanae based on seven specimens from a relatively small geographical region in Otago and northern Fiordland. We have expanded the known range for the species considerably to the north, south and east (Fig. 18B). While specimens collected along the length of the Southern Alps and Fiordland are morphologically indistinguishable, they fall into three genetically and geographically distinct clades. Specimens collected in the mountains either side of Lake Wakatipu and south of the Greenstone and Kawarau River show some noticeable morphological differences compared to the populations further west. We thus propose to recognise four separate subspecies. (Karny, 1935) comb. nov. Figs 1-2, 5-6, 8D, 9B, 10D-E, 12D-F, 14C-D, 16A-C, 18B

Diagnosis
A mid-sized cave wētā found in the Southern Alps from Aoraki/Mt Cook south to the Dart River and east into Central Otago, at elevations greater than 1200 m. Body colour varies from chequered yellow/black to all black. Dorsal surfaces may be covered in fine tomentum, giving the insect a matt appearance, or glabrous, giving the insect a shiny appearance. The latter trait is more common at the northern end of the species' distribution range.
At the northern end of its distribution range, P. cochleatus cochleatus overlaps with P. montanus and can easily be confused with the latter, especially at the higher elevations, where both insects are entirely black. The most reliable trait to differentiate the two species is the presence of dorsal spines on the first hind tarsus segment in P. cochleatus.
Across much of its range, P. cochleatus cochleatus overlaps with either Notoplectron brewsterense comb. nov. or with Talitropsis chopardi (Karny, 1937) east of the Main Divide of the Southern Alps. Both of the latter two species have lighter colour, shorter legs with fewer spines on the hind tibiae and no dorsal spines on the first hind tarsus segment.

Etymology
'Cŏchlĕăr' is Latin for 'spoon'. 'Cochleatus' means 'equipped with a spoon' -after the shape of the male subgenital plate.
Material examined (see also Supp. file 1:    Head. As per generic description. Eyes dark. Specimens living above the permanent snowline are entirely black. This includes all head parts that are usually pale otherwise, e.g., maxillary palps and antennae.

THorax. As per generic description.
Legs. Notably shorter than in P. montanus. The hind tibiae are approx. 10% shorter than the body in both males and females. Fore femora always unarmed at the apex. Mid femora armed with one retrolateral spine at the apex; prolateral apical spine absent. The first hind tarsus segment is armed with a variable number of linear spines above (up to nine); the second hind tarsus segment is only rarely armed with dorsal linear spines. The first tarsal segment on all six legs is pale at times, but not always; the insect can be entirely black (Fig. 14D).
abdomen. Tergites tomentose or glabrous; the latter trait appears to be more common in the northern half of the range. A thin, pale median line along the whole length of the insect may or may not be present. The colour of the tergites is mostly chequered yellow/black or light grey-brown/black at the lower elevations (Fig. 14C), transitioning to mostly black (Fig. 8D) or all black (Fig. 14D) at the higher elevations, especially above the permanent snow-line. maLe TerminaLia. Subgenital plate an isosceles triangle with very rounded corners and a spoon on the vertex; keeled and mostly glabrous, with some sparse hair near the edges; shallow when seen from the side. In a dorsal view, only the spoon is visible, protruding beyond the paraprocts; these are covered in short, dense bristles ( Fig. 10D-F).
FemaLe TerminaLia. Subgenital plate bilobed, the two rounded lobes short, broad, and asymmetrical, with a wide gap in between; very hairy ( Fig. 12D). Ovipositor on average 70% of body length, only very gently curving upwards; lower valve with 5 to 9 strong teeth below at the apex ( Fig. 12E-F Pharmacus cochleatus nauclerus is sympatric with Pharmacus cristatus sp. nov. and with Notoplectron brewsterense comb. nov. It is easily differentiated from the former by its darker coloration and lack of prolateral apical spines on the fore and mid femora and from the latter by its darker colour, longer legs with more numerous spines on the hind tibiae and the presence of dorsal spines on the first hind tarsus segment.

Etymology
'Nauclērus' is Latin for 'skipper', after the Skippers Range in South Westland.
Material examined (see also Supp. file 1: In Fiordland, Pharmacus cochleatus fiordensis subsp. nov. is sympatric with Pharmacus cristatus sp. nov. It is easily differentiated from the latter by its darker coloration and lack of prolateral apical spines on the fore and mid femora.
Across much of its range, P. cochleatus fiordensis overlaps with Notoplectron brewsterense comb. nov. The latter species has lighter colour, shorter legs with fewer spines on the hind tibiae and no dorsal spines on the first hind tarsus segment.

Etymology
'Fiordensis', inhabits Fiordland, the geographical region in the south-west corner of New Zealand. Crowe (2002) refers to this insect as 'black tumbling cave wētā' in light of his observations of how "to escape danger, it leaps, then rolls down the scree to tumble into a gap between the stones".
Material examined (see also Supp. file 1:  Pharmacus cochleatus fiordensis subsp. nov. is morphologically almost identical to Pharmacus cochleatus cochleatus comb. nov. See the description for the latter. While Pharmacus cochleatus fiordensis subsp. nov. is a larger insect on average and tends to have a longer ovipositor in relation to body length, the overlap between the two subspecies' physical dimensions is such that measurements cannot be used to discriminate between them.

Diagnosis
A mid-sized cave wētā found in the mountains either side of Lake Wakatipu and south of the Greenstone and Kawarau Rivers, at elevations greater than 1300 m. Body colour is chequered light brown/dark brown, with reddish legs and antennae. A prolateral apical spine is often present on the fore femur, and occasionally on the mid femur also. Hind legs longer than in P. cochleatus cochleatus.
The sympatric Pharmacus notabilis sp. nov. is mainly found at lower elevations and is easily identified by its turquoise/grey tergites, as well as its unique male terminalia. Two other species of Rhaphidophoridae share the habitat with P. cochleatus rawhiti subsp. nov., these are Talitropsis chopardi and Macropathus sp. . The former has shorter legs and fewer, larger spines on the hind tibiae, and unarmed hind tarsi; the latter has uniform brown colour and much longer legs, which give the insect a very slender appearance.

Etymology
'Rāwhiti' is Te Reo for 'eastern' -this subspecies inhabits mountain ranges east of the Southern Alps.
Unaware of the fact that it belongs to the same species as his 'black tumbling cave wētā', Crowe (2002) refers to this insect as 'Remarkables cave wētā'. We discourage the use of this common name, since there are at least four species of cave wētā found in The Remarkables, including two species of Pharmacus. The range of Pharmacus cochleatus rawhiti subsp. nov. extends well beyond The Remarkables, as far south as the Garvie and Umbrella Mountains, and west to the Livingstone Mountains.
Material examined (see also Supp. file 1: Head. As per generic description. Eyes grey/green. Antennae red-brown.

THorax. As per generic description.
Legs. Longer than in P. cochleatus cochleatus; approx. same length as in P. montanus. The hind tibiae are on average 5% longer than body in females and 20% longer than body in males. Fore femora often armed with one prolateral spine at the apex. Mid femora always armed with one retrolateral spine and occasionally with one prolateral spine at the apex. The first hind tarsus segment is armed with a variable number of dorsal linear spines above.
abdomen. Tergites sparsely tomentose. A thin, pale median line along the whole length of the insect may or may not be present. The colour of the tergites is mostly chequered light/dark red-brown or brown (Fig. 8E). Nymphs may be uniform brown.

Diagnosis
A mid-sized cave wētā found in the alpine regions of northern Fiordland and South Westland, at elevations greater than 1000 m. Body colour is chequered pale/brown, with variegated legs and brown antennae. A prolateral apical spine is always present on both fore and mid femora.
The species is sympatric with P. cochleatus nauclerus subsp. nov. in the Skippers Range (South Westland) and with P. cochleatus fiordensis subsp. nov. in Fiordland. Across its whole distribution range, it is also sympatric with Notoplectron brewsterense comb. nov. It is differentiated from all of these taxa by the presence of prolateral apical spines on fore and mid femora.

Etymology
'Cristātus' is Latin for 'tufted'. Named after the prominent hair tufts on the vertex of the male subgenital plate (Fig. 10G-H).  (Karny, 1935)  Head. As per generic description. Eyes grey/brown. Antennae brown.

THorax. As per generic description.
Legs. Relative to the body, longer than in any other Pharmacus species. The hind tibiae are on average 15% to 20% (max. 45%) longer than the body in both males and females. Fore and mid femora always armed at the apex with a prolateral spine, in addition to the retrolateral apical spine on the mid femur. The first hind tarsus segment is armed with a variable number of dorsal linear spines.
abdomen. Tergites sparsely tomentose. A thin, pale median line along the whole length of the insect may or may not be present. The colour of the tergites is mostly chequered pale/brown (Fig. 8F). maLe TerminaLia. Subgenital plate similar to P. cochleatus, but shorter and bent upwards. Moderately hairy underneath, with two characteristic hair tufts on the spoon's vertex, both dorsal and ventral ( Fig. 10G-I).
FemaLe TerminaLia. Very similar to P. cochleatus. Subgenital plate bilobed, the two rounded lobes short, broad, and asymmetrical, with a wide gap in between; very hairy ( Fig. 12G). Ovipositor on average 70% of body length, only very gently curving upwards; lower valve with 5 to 9 strong teeth below at the apex ( Fig. 12H-I).

Diagnosis
A mid-sized cave wētā found in the low alpine regions of The Remarkables and Hector Mountains, east of Lake Wakatipu, at elevations between 1200 m and 1600 m. Body colour is chequered turquois/ grey and eye colour is blue/grey. A prolateral apical spine on the fore femur is always present in males, occasionally in females.
The sympatric Pharmacus cochleatus rawhiti subsp. nov. is mainly found at higher elevations and is easily differentiated by its brown tergites and reddish legs and antennae. Two other species of Rhaphidophoridae share the habitat with P. notabilis sp. nov., these are Talitropsis chopardi and Macropathus sp. . The former has shorter legs and fewer, larger spines on the hind tibiae, and unarmed hind tarsi; the latter has uniform brown colour and much longer legs, which give the insect a very slender appearance.

Etymology
'Nŏtābĭlis' is Latin for 'remarkable', after the location where the species was first discovered, The Remarkables near Queenstown. Pharmacus notabilis means 'remarkable sorcerer'.

Holotype
Head. As per generic description. Eyes grey/blue. Antennae brown.
THorax. As per generic description.
Legs. The hind tibiae are approx. 10% Shorter than the body in females, equally long as body in males. Fore femora always armed at the apex with one prolateral spine in males, occasionally also in females. Mid femora armed with one retrolateral spine at the apex; prolateral apical spine absent. The first hind tarsus segment is armed with a variable number of linear spines above (up to eight); the second hind tarsus segment is only rarely armed with dorsal linear spines.
abdomen. Tergites glabrous or sparsely tomentose (Fig. 8H), coloured chequered brown in nymphs, chequered turquois/grey in adults (Fig. 15C). A thin, pale median line along the whole length of the insect is generally present. maLe TerminaLia. Subgenital plate unmistakeable, elongated, the vertex of the spoon acuminate, shallow when seen from the side (Fig. 10J-L). FemaLe TerminaLia. Subgenital plate small, narrow at the base and widening at the vertex, bilobed, the two rounded, narrow lobes very hairy (Fig. 12J). Ovipositor relatively short, on average 60% of body length, gently curving upwards; lower valve with 5 to 7 teeth below at the apex (Fig. 12K-L).

Diagnosis
A mid-sized cave wētā known only from the rock tors on the wind-swept tops of the Old Man, Old Woman and Dunstan Ranges in Central Otago. Typical of Pharmacus in its habitus, it is differentiated from all other species by its short hind legs. It does not share the habitat with any other Rhaphidophoridae that we are aware of.

'Senex' is Latin for 'old' (of a person), after the Old Man and Old Woman Ranges in Central Otago.
Pharmacus senex means 'old sorcerer'.

Holotype
Head. As per generic description. Eyes grey or brown. Antennae brown.
THorax. As per generic description.
Legs. All legs coloured a variegated reddish brown in adults, dull in nymphs. Hind tibiae approx. three quarters of body length in females; just over 80% of body length in males. This makes the hind legs of P. senex the shortest (proportionally to body size) in all Pharmacus species (Fig. 9F). Fore femora generally unarmed at the apex. One specimen (female) out of 29 examined had a prolateral spine at the apex of the fore femur. Mid femora armed with one retrolateral spine at the apex; prolateral apical spine absent. The first hind tarsus segment is armed with a variable number of linear spines above (up to seven); the second hind tarsus segment is only rarely armed with dorsal linear spines.
abdomen. Tergites glabrous or sparsely tomentose, coloured chequered a vibrant yellow/brown in adults (Fig. 15A-B), dull brown in nymphs. A thin, pale median line along the whole length of the insect is generally present. maLe TerminaLia. Subgenital plate deep like in P. montanus when seen from the side, but not covered by a membrane in the distal half; square-ended and hairy at the vertex. Styli long, reaching almost to the end of the subgenital plate ( Fig. 11A-C).
FemaLe TerminaLia. Subgenital plate bilobed and wide, similar to the one in P. cochleatus comb. nov. (Fig. 13A). Ovipositor on average 70% of body length, nearly straight; lower valve with 5 to 7 teeth below at the apex (Fig. 13B-C).

Diagnosis
A mid-sized cave wētā known only from the alpine regions of the Eyre Mountains south of Lake Wakatipu; it is the largest of all Pharmacus species. Body colour is dark, nearly black.
It shares the habitat with Talitropsis chopardi and Macropathus sp. The former has lighter colour, shorter legs and fewer, larger spines on the hind tibiae, and unarmed hind tarsi; the latter has uniform brown colour and much longer legs, which give the insect a very slender appearance.
The species cannot be reliably differentiated from Pharmacus cochleatus comb. nov. without an examination of the terminalia. The two species however are not sympatric.
Material examined (see also Supp. file 1: Head. As per generic description. Eyes grey or brown. Antennae brown or black.
THorax. As per generic description.
Legs. Variegated colour, occasionally dark or almost black; hind tibiae black above. Hind tibiae approximately as long as body, or just under, in both males and females. Fore femora armed with one prolateral spine at the apex in four out of twelve females and one out of ten males examined. Mid femora armed with one retrolateral spine at the apex; prolateral apical spine absent. The first hind tarsus segment is armed with a variable number of linear spines above (up to ten); the second hind tarsus segment is only rarely armed with dorsal linear spines.
abdomen. Tergites glabrous or sparsely tomentose; colour chequered yellow/black in adults (Fig. 15D), the black being predominant and giving the insect a dark look. A thin, pale median line along the whole length of the insect is generally present.
maLe TerminaLia. Subgenital plate deep when seen from the side; shaped like a spoon when seen from above, broader at the vertex and less constricted at the neck compared to the subgenital plate of Pharmacus cochleatus comb. nov. Paraprocts large and swollen, bristled on the inside, glabrous on the outside (Fig. 11D-F).
FemaLe TerminaLia. Subgenital plate bilobed, the incision between the two lobes sutured (Fig. 13D), a trait not found in any other Pharmacus species. Ovipositor long and straight, just under three quarters of body length; lower valve with 6 to 8 teeth below at the apex (Fig. 13E-F The species shares the habitat with Talitropsis chopardi. The latter is larger, has fewer, larger spines on the hind tibiae, has unarmed hind tarsi, and generally shows a prominent dorsal yellow stripe or diamond on the abdomen in the Takitimu Mountains population.

Etymology
'Perfĭdus' (noun) is Latin for 'criminal', 'crook'. Pharmacus perfidus means 'criminal sorcerer'. Try scrambling on the jagged, crumbly ridgelines of the Takitimu Mountains at night and you will know why. It is also banished to an isolated mountain range, a great distance away from all other Pharmacus species (see Fig. 6).
Material examined (see also Supp. file 1: Head. As per generic description. Eyes grey/brown. Antennae brown. Frons dark brown or black; darker than body.
THorax. As per generic description.
Legs. Variegated colour, mostly brown; hind tibiae dark above. Hind tibiae approximately as long as body in males, 10% shorter than body in females. Fore femora armed with one prolateral spine at the apex in four out of five females and seven out of nine males examined. Mid femora armed with one retrolateral spine at the apex; prolateral apical spine absent. The first hind tarsus segment is armed with a variable number of linear spines above (up to seven); the second hind tarsus segment is only rarely armed with dorsal linear spines.
abdomen. Tergites tomentose; colour chequered light/dark brown (Figs 8J, 15E). A thin, pale median line along the whole length of the insect is generally present. maLe TerminaLia. Nearly identical to the male terminalia of Pharmacus concinnus sp. nov. Subgenital plate deep when seen from the side; shaped like a spoon when seen from above, broader at the vertex and less constricted at the neck compared to the subgenital plate of Pharmacus cochleatus comb. nov. Paraprocts large and swollen, bristled on the inside, glabrous on the outside (Fig. 11G-I).
FemaLe TerminaLia. Subgenital plate bilobed, the two lobes narrow and elongated (Fig. 13G), very similar to the subgenital plate of Pharmacus montanus. Ovipositor longer in proportion to body size than in any other Pharmacus species; approximately 80% of body length; lower valve with 6 to 8 teeth below at the apex (Fig. 13H-I).

Diagnosis
A mid-sized cave wētā known only from the low to mid alpine regions at the foot of Mount Aspiring in the Southern Alps. It is one of the smaller Pharmacus species. Body colour is chequered brown. The fore femur is always armed with a prolateral apical spine.
The species is sympatric with Notoplectron brewsterense comb. nov. and with Pharmacus cochleatus cochleatus comb. nov., although it occupies lower elevations compared to the latter. Notoplectron brewsterense has fewer, larger spines on the hind tibiae and has unarmed hind tarsi. Pharmacus cochleatus cochleatus is larger, has darker colour, and lacks the prolateral apical spine on the fore femur.

Etymology
'Vallestris' is Latin for 'of the valley'. In contrast to sympatric species that inhabit the mountain tops.
Material examined (see also Supp. file 1: Head. As per generic description. Eyes brown/black. Antennae brown.
THorax. As per generic description.
Legs. Variegated colour, mostly brown. Hind tibiae of equal length as body or longer in both males and females. Fore femora armed with one prolateral spine at the apex. Mid femora armed with one retrolateral spine at the apex; prolateral apical spine absent. The first hind tarsus segment is armed with a variable number of linear spines above (up to seven); the second hind tarsus segment is only rarely armed with dorsal linear spines.
abdomen. Tergites tomentose; colour chequered light/dark brown (Figs 8G, 15F). A thin, pale median line along the whole length of the insect is generally present. maLe TerminaLia. Subgenital plate roughly in the shape of a French wine cask, similar to but shorter than in Pharmacus cochleatus comb. nov.; very thin and translucent; the paraprocts visible through the subgenital plate from underneath ( Fig. 11J-L).
FemaLe TerminaLia. Subgenital plate bilobed, the two rounded lobes short, broad, and asymmetrical, with a wide gap in between (Fig. 13J); very similar to the subgenital plate of Pharmacus cochleatus comb. nov. Ovipositor relatively short and curved upwards; on average 65% of body length; lower valve with 6 teeth below at the apex (Fig. 13K-L).

Habitat and distribution
Rhaphidophorids in the genus Pharmacus are true alpine specialists. They are only found above the tree line, at elevations of 1300 m a.s.l. or higher and well into the nival zone. Our lowest record of Pharmacus is a male P. cristatus sp. nov. collected at 1000 m a.s.l. at the head of Mistake Creek, Earl Mountains ( Richards (1972 The genus Pharmacus is endemic to the South Island of New Zealand. Pharmacus montanus and P. cochleatus are the two most widespread species, occupying the length of the Southern Alps from Fiordland to Kahurangi National Park between them. Pharmacus cochleatus is found in the south from Fiordland to Aoraki/Mt Cook, while P. montanus is found in the northern half of the Southern Alps from Aoraki/Mt Cook to Mt Owen in Kahurangi National Park. Our southern-most specimen of P. montanus was collected on Mt Dark in the Ben Ōhau Range (43.871° S, 170.045° E), whereas our northern-most specimen of P. cochleatus is from Mt Edgar Thomson (43.778° S, 170.064° E), also in the Ben Ōhau Range, 12 km north of Mt Dark. These locations pin-point a narrow region just south of Aoraki/Mt Cook where the two species overlap (Fig. 1B). A similar geographical and phylogenetic divide either side of the Mt Cook region had been previously documented for numerous alpine birds and insects (Weston & Robertson 2015;Wallis et al. 2016) and is attributed to extensive glacial ice during the Early Pleistocene glaciation 1.8 million years ago (Wallis et al. 2016). This barrier might also have separated Pharmacus montanus and Pharmacus cochleatus 2 Ma ago.
Additional genetic and morphological differentiation has occurred within Pharmacus cochleatus in the southern regions, justifying the recognition of four subspecies. The divide between Pharmacus cochleatus cochleatus and Pharmacus cochleatus fiordensis subsp. nov. follows a line along the major glacial axis of the Dart River and Lake Wakatipu. Pharmacus cochleatus rawhiti subsp. nov. spans Lake Wakatipu and is delimited by the low-lying Greenstone and Eglinton Rivers to the north and west, and by the Kawarau River to the north and east. Pharmacus cochleatus nauclerus subsp. nov. appears to be endemic to the Skippers Range in South Westland, a true alpine island surrounded by land entirely below 50 m a.s.l. The lack of mtDNA COI sequence differentiation between Pharmacus cochleatus fiordensis and Pharmacus cochleatus rawhiti may be explained by recent connectivity of the populations on either side of the Greenstone River. However, it is not uncommon for mtDNA (DNA barcoding) to be unable to discriminate recently divergent lineages due to incomplete lineage sorting or hybridisation (e.g., polar bears - Liu et al. 2014;bison -Wang et al. 2018;Drosophila -Yassin et al. 2010).
Pharmacus cristatus sp. nov. is currently known from the Skippers Range in South Westland and from the Darran and Earl Mountains in Fiordland, west of the major depression in the landscape aligned with the Eglinton, Hollyford and Pyke Rivers (Hollyford Fault System). Pharmacus cristatus sp. nov. is sympatric with P. cochleatus throughout its range; the two species share the same habitat and form mixed populations. However, not only are they different enough morphologically to be easily identified in the field, they also differ in their patterns of activity. While primarily nocturnal, Pharmacus cochleatus is often seen out and about during daylight and jumps readily when disturbed by the light of a torch. P. cristatus on the other hand appears to be exclusively nocturnal and freezes when exposed to a source of light.
While west of the Eglinton River we have two sympatric species, Pharmacus cochleatus fiordensis subsp. nov. and P. cristatus sp. nov., east of the Eglinton and south of the Greenstone River there is only one species, P. cochleatus rawhiti subsp. nov. The latter has the terminalia and mitochondrial DNA of P. cochleatus fiordensis, but its brown body colour, apical spine combination and the longer legs are more similar to P. cristatus sp. nov. Whether this is due to gene flow or convergent evolution would require information from additional genetic markers.
The remaining five species of Pharmacus appear to be restricted to much smaller geographical regions. Pharmacus perfidus sp. nov. and Pharmacus concinnus sp. nov. are endemic to the Takitimu and the Eyre Mountains respectively, two well defined alpine islands. Pharmacus senex sp. nov. is currently only known from the Old Man and Old Woman Ranges and from the Dunstan Mountains in Central Otago. None of these mountain ranges are geographically isolated. Pharmacus senex sp. nov. appears to be defined by habitat rather than geographical isolation. It is only found on the rock tors that dot the desolated high plateaus of Central Otago. The extreme winds that constantly buffet these high plateaus are a plausible explanation for the species' short legs, since opportunities to jump are limited and long legs could in fact be detrimental, increasing the chance of the insect being caught and swept away by the wind.
Pharmacus notabilis sp. nov. appears to be confined to The Remarkables and Hector Mountains south of Queenstown, where it is sympatric with P. cochleatus rawhiti subsp. nov. The two species seem to occupy different elevation bands, P. notabilis sp. nov. at elevations lower than approx. 1500 m a.s.l., P. cochleatus rawhiti subsp. nov. at the higher elevations. So far, we have not found any populations where the two species mix.
Pharmacus vallestris sp. nov. is currently only known from the Matukituki River West Branch, where it is sympatric with Pharmacus cochleatus cochleatus, but inhabits lower elevations compared to the latter. The two species seem to utilize different habitats, too, as P. vallestris sp. nov. was found in screes and small boulders, whereas the habitat where P. cochleatus cochleatus was found mostly consists of fissures in solid rock. However, our number of P. vallestris sp. nov. observations is small. The seemingly limited range of P. vallestris sp. nov. is somewhat puzzling. We searched the adjacent Matukituki River East Branch and the nearby Earnslaw Burn at different elevations and in different habitats, but only found Pharmacus cochleatus cochleatus. We consider it most likely that P. vallestris sp. nov. may exist cryptically in neighbouring valleys and expect future searches to extend the species' known distribution range.
We note that Pharmacus notabilis sp. nov. and P. vallestris sp. nov. share a recent common ancestor (see Fig. 2) and are both 'low elevation' species. We speculate that separation between the species groups may have occurred during a glacial maximum, Pharmacus cochleatus surviving at high elevation on nunataks, P. notabilis sp. nov. and P. vallestris sp. nov. below the glaciers. All species would have expanded their range in warmer times, top-down in the case of Pharmacus cochleatus, bottom-up in the case of P. notabilis sp. nov. and P. vallestris sp. nov. In the current interglacial, Pharmacus cochleatus, P. montanus and Notoplectron brewsterense comb. nov. are the only rhaphidophorids found on nunataks in the glaciated regions of the Southern Alps today.
New Zealand glaciers are currently in a phase of rapid recession (Willsman et al. 2015) as a result of anthropogenic climate change (Vargo et al. 2020), and alpine invertebrates are responding to the shifting snow-line by moving to higher elevations (Chinn & Chinn 2020). Pharmacus montanus and Pharmacus cochleatus cochleatus occupy elevation bands that are at least 1000 m wide and have plenty of room to move upwards. Pharmacus perfidus sp. nov., P. senex sp. nov. and P. cochleatus nauclerus subsp. nov. on the other hand are confined to elevation bands less than 300 m wide on mountain ranges with a maximum elevation of 1650 m a.s.l. These species and subspecies are at risk of extinction as their alpine islands are 'submerged' as a consequence of global warming.

Notes on the biology of Pharmacus
Pharmacus take shelter in fissures in rock cliffs and ridges and in holes under boulder fields. While they are predominantly nocturnal, in the nival zone of the Southern Alps they are often active during the day and are seen by climbers on hot, sunny days as they jump from rock onto snow (hence the common name 'Mount Cook flea'). We propose two possible explanations for diurnal activity in the high alpine regions. Firstly, night-time temperatures at high elevation frequently drop well below 0°C even in summer, making foraging difficult, especially when lichens become coated in ice. Milder day-time temperatures may thus present better conditions for foraging. Secondly, on hot, sunny days Pharmacus may seek snow to cool down and avoid desiccation, as solar radiation can warm up the rock to temperatures that feel uncomfortable on climbers' fingers, conditions that are probably unbearable for insects.
Pharmacus are mainly herbivore, although they probably also scavenge on other invertebrates when available. Dumbleton (1952) noted that "examination of faecal pellets from a specimen from Mt Mahanga suggested that insect flotsam and jetsam on the snow may form at least part of their diet". Sweney (1980) also examined faeces of Pharmacus montanus in Aoraki Mt Cook National Park and found that they only contained plant fragments. Sweney (1980) observed P. montanus being associated with Veronica haastii Hook.f., Veronica birleyi N.E.Br., Chionochloa spp., Dolichoglottis scorzoneroides (Hook.f.) B.Nord., Ranunculus grahamii Petrie and rock lichens. The vast majority of Pharmacus we have observed in the wild were feeding on rock lichen, only occasionally on Chionochloa spp. We have never observed Pharmacus actively scavenging on other invertebrates. We also consider it likely that when on snow, Pharmacus may feed on the rich red algal growth that coats snow-fields during the summer season (Novis 2002), a condition commonly known as 'watermelon snow' or 'red snow'.
The life span and the number of instar stages in Pharmacus are unknown. Richards (1961) found nine pre-adult instars in male and seven in female Pachyrhamma edwardsii (Scudder, 1869) and a duration of two years for the complete life cycle of Pachyrhamma cavernae (Hutton, 1900); both species are larger than Pharmacus. Eight male and nine female pre-adult instars occur in Insulanoplectron spinosum (Butts 1983, while there are seven in both males and females of the Australian rhaphidophorid Pallidotettix nullarborensis Richards, 1968. Studies on cave dwelling Rhaphidophoridae overseas have shown a lifespan of one year in Ceuthophilus stygius (Scudder, 1861) (Lavoie et al. 2007), two and a half years in Dolichopoda geniculata (Costa, 1860) (Carchini et al. 1994). Similar studies on high alpine Rhaphidophoridae are lacking.
Given the short duration of the snow-free season on New Zealand mountains (November to April), it is unlikely that Pharmacus could complete their life cycle within one season. Pharmacus adults and nymphs are active in spring as soon as temperatures become warm enough (pers. obs.), indicating that these insects can survive winter in a dormant state at all life stages. This life history trait is also typical of New Zealand's alpine grasshoppers (family Acrididae MacLeay, 1821) (Ramsay 1978;Koot et al. 2020). Unpredictable seasonal conditions in the New Zealand alpine zone may result in uneven development rates and growth whenever the environment permits (Sutherland 1964), which is consistent with the prevalence of freeze tolerance in southern hemisphere insects (Sinclair et al. 2003;Sinclair & Chown 2005). Freeze tolerance is known in other New Zealand Orthoptera, including species of Deinacrida White, 1842 and Hemideina Walker, 1869 in the family Anostostomatidae de Saussure, 1859 (Sinclair et al. 1999(Sinclair et al. , 2003Wharton 2011). Pharmacus however shelter deep inside rock crevices, where temperature fluctuations are reduced by the rocks' thermal inertia and in winter by the additional presence of snow cover. Once the ground freezes over for winter, we consider it unlikely that they would resume any activity before spring. From this point of view, Pharmacus are subject to environmental conditions more typical of the northern hemisphere. We have observed Pharmacus actively forage at air temperatures as low as -5°C, although they move slowly and are unable to jump. The ability of Pharmacus to remain active at air temperatures below 0°C, combined with the fact that they live at elevations much higher than any New Zealand invertebrates known to be freeze-tolerant, suggests that in Pharmacus freeze avoidance may play a greater role than freeze tolerance (Sinclair & Chown 2005).
Melanism in Pharmacus is apparent where they occur at the higher elevations. On the Sealy Range in Aoraki/Mt Cook National Park, Pharmacus montanus is brown at 1500 m a.s.l. (Fig. 14B), mottled brown/black at 1700 m to 1800 m a.s.l. (pers. obs.) and entirely black at 2200 m a.s.l. (Fig. 14A). Throughout the bottom half of the Southern Alps, Pharmacus cochleatus is chequered yellow/black at 1500 m a.s.l. (Fig. 14C), transitioning to pitch black at higher elevation (Fig. 14D). Melanism is likely to provide a two-fold advantage in the high alpine regions where air temperature is low but solar radiation is high. Firstly, darker individuals are expected to warm up faster when exposed to sun-light, resulting in increased ability to jump and escape predation. This would be especially important for Pharmacus in the nival zone where they are commonly diurnal. The thermal melanism hypothesis has been tested in alpine grasshoppers in the European Alps (Köhler et al. 2017) and in Australia (Harris et al. 2013), and in several species of winged insects (Roland 1982;Guppy 1986;Wittkopp et al. 2011). Secondly, cuticular melanism is associated with increased resistance to desiccation in several invertebrate taxa (Parkash 2010) including the mountain stone wētā Hemideina maori (Pictet & de Saussure, 1893) (King & Sinclair 2015).
In the Southern Alps, Pharmacus and Notoplectron brewsterense comb. nov. are sympatric and are often found in mixed populations on the same cliffs. At medium elevations (1500 m to 1800 m) Pharmacus are outnumbered by Notoplectron at a ratio of 10:1 or higher. Indeed, Notoplectron brewsterense comb. nov. at times form dense colonies, emerging onto alpine cliffs after dark in their hundreds. It is not only a very widespread but also a very common species. At the higher elevations (above 2000 m) Pharmacus seems to be more common, but we lack enough data to support our sparse anecdotal evidence. East of the Southern Alps, in the mountains of Southland and Otago, Notoplectron brewsterense comb. nov. is replaced by Talitropsis chopardi. While Pharmacus and Talitropsis chopardi are also sympatric and found in mixed populations, here the ratios are inverted, as Pharmacus are by far the most numerous and form dense colonies. It would seem that, while Pharmacus are highly adapted to life in the most inhospitable alpine habitats, the more sheltered ranges of Southland and Otago are where conditions are optimum for these cave wētā to thrive. In the rain-parched ranges of East Canterbury and Marlborough however, Pharmacus are entirely absent and are replaced by Rhaphidophoridae in the genus Petrotettix Richards, 1972.

Morphology of Pharmacus and differentiation from other New Zealand Rhaphidophoridae
Pharmacus is a well-defined, compact genus of Rhaphidophoridae. The exclusively alpine habitat, ovipositor with smooth upper valve, triangular male subgenital plate with a 'spoon' at the apex, body length between 10 mm and 18 mm and the slender legs with two rows of 20 to 30 dorsal linear spines on the hind tibiae differentiate it from all other New Zealand Rhaphidophoridae. Some interesting considerations arise from an examination of the apical spines on the fore and mid femora of Pharmacus (Table 1). In his synopsis of the genera of New Zealand Rhaphidophoridae, Hutton (1896) assumed the presence of prolateral and retrolateral spines at the apex of each femur to be characteristic of a genus. The presence of a retrolateral spine at the apex of the fore femur in Pleioplectron diversum Hutton, 1896 for instance was enough for Hutton (1898) to justify moving the species to a new genus, Miotopus Hutton, 1898. His view on the importance of the spines at the apex of each femur remained unchallenged for decades. Hutton (1896) described Pharmacus as having "all femora without apical spines".
Based on Hutton's previous work, Karny assigned his new species Isoplectron cochleatum to the genus Isoplectron Hutton, writing that "in light of its body build and spines on the legs this species belongs without doubt to the genus Isoplectron". In more recent times, the only two published keys to the genera of New Zealand Rhaphidophoridae still assume the configuration of spines at the apex of each femur to be a key characteristic of a genus (Ward 1997;Johns & Cook 2013). Richards (1959) was first to challenge the idea when dealing with the species Miotopus diversus (Hutton, 1896), writing that "Admittedly the presence of two apical spines on the fore femora is a different character to that found in Pleioplectron; but variation is so common that too much weight cannot be placed on it when only one specimen is available for examination". And yet Richards defined a number of new species based on the configuration of apical spines of the material she examined; in her generic description of Pharmacus (1972) she wrote "Apical spines on legs constant in number. Fore femur unarmed; fore tibia bears four apical spines, one above and one beneath both prolaterally and retrolaterally; fore tarsus unarmed. Middle femur bears one prolateral apical spine beneath; middle tibia bears four apical spines, one above and one beneath both prolaterally and retrolaterally; middle tarsus unarmed. Hind femur unarmed".
In their work on the genus Pachyrhamma Brunner von Wattenwyl, 1888, Cook et al. (2010) found that the configuration of spines at the apex of the hind femora varies not only within the genus, but also within a species. The same applies to the genus Pleioplectron Hudson, 1896 (Hegg et al. 2019). The presence of a spine at the apex of the hind femur can thus not be used as a trait to characterise a species. Hegg et al. (2019) also found that a retrolateral spine at the apex of the fore femur is present in four out of ten species in the genus Pleioplectron. The configuration of apical spines on the fore and mid femur however is constant within each species.
In the genus Pharmacus, only three out of eight species have a constant configuration of apical spines on the fore and mid femora (see Table 1). These are Pharmacus montanus (27 specimens examined), P. cristatus sp. nov. (11 specimens examined) and P. vallestris sp. nov. (6 specimens examined). Larger sample sizes could introduce variation in some of these species also. In Pharmacus cochleatus comb. nov., three out of four subspecies also have a constant configuration of apical spines on the fore and mid femora: P. cochleatus cochleatus (56 specimens examined), P. c. nauclerus subsp. nov. (10 specimens examined) and P. c. fiordensis subsp. nov. (8 specimens examined). In all other species or subspecies, the only constants are the presence of a retrolateral spine at the apex of the mid femur, and the absence of a retrolateral spine at the apex of the fore femur. A prolateral spine at the apex of the fore and/or mid femur may or may not be present. In Pharmacus cochleatus rawhiti subsp. nov., we have found three different configurations of apical spines on the fore and mid femora of ten specimens collected on the same rock.
In general, the configuration of apical spines on the fore and mid femora of New Zealand Rhaphidophoridae is a useful trait when disambiguating between species and even genera. However, there are exceptions; the genus Pharmacus is one of these exceptions. We do not know if apical spines on the fore and mid femora of Rhaphidophoridae are functional, or what their function might be. We highlight the fact that the configuration of spines at the apex of the femora in Rhaphidophoridae cannot be taken as sole evidence to define a species, let alone a genus. This is consistent with the findings from previous studies on the European genus Dolichopoda Bolívar, 1880, which showed that spinulation in Rhaphidophoridae is not a valid trait for resolving systematic issues (Allegrucci et al. 2005).

A note on nomenclature
In reassigning the species Pharmacus brewsterensis to the genus Notoplectron we have assumed Notoplectron to be gender neuter. This is in spite of Richards (1964) having named the only existing species in the genus Notoplectron campbellensis.
of Latin declensions when naming Notoplectron campbellensis Richards, 1964 and Dendroplectron aucklandensis Richards, 1964. In compliance with article 31.2 of the ICZN Code (ICZN 1999), the correct declension for the Auckland Islands cave wētā is Dendroplectron aucklandense Richards, 1964 and the correct declension for the Campbell Island cave wētā is Notoplectron campbellense Richards, 1964. Likewise, the correct declension for the new combination of Pharmacus brewsterensis is Notoplectron brewsterense (Richards, 1972) comb. nov.